3-dimensional NOR memory array architecture and methods for fabrication thereof
A method addresses low cost, low resistance metal interconnects and mechanical stability in a high aspect ratio structure. According to the various implementations disclosed herein, a replacement metal process, which defers the need for a metal etching step in the fabrication process until after all patterned photoresist is no longer present. Under this process, the conductive sublayers may be both thick and numerous. The present invention also provides for a strut structure which facilitates etching steps on high aspect ratio structures, which enhances mechanical stability in a high aspect ratio memory stack.
Latest SunRise Memory Corporation Patents:
- Three-dimensional memory string array of thin-film ferroelectric transistors
- Methods for forming multi-layer vertical NOR-type memory string arrays
- Quasi-volatile system-level memory
- Staircase structures for electrically connecting multiple horizontal conductive layers of a 3-dimensional memory device
- Semiconductor memory device with write disturb reduction
This application is a continuation application of U.S. patent application (“Parent Application”), Ser. No. 17/011,836, entitled “3-Dimensional NOR Memory Array Architecture and Methods for Fabrication Thereof,” filed on Sep. 3, 2020, which is a continuation of U.S. patent application Ser. No. 16/792,790, entitled “3-Dimensional NOR Memory Array Architecture and Methods for Fabrication Thereof,” filed on Feb. 17, 2020, which is a continuation of U.S. patent application Ser. No. 16/012,731, entitled “3-Dimensional NOR Memory Array Architecture and Methods for Fabrication Thereof,” filed on Jun. 19, 2018, which claims priority of: (i) U.S. provisional application (“Provisional Application I”), Ser. No. 62/522,666, entitled “Replacement Metal and Strut for 3D memory Array,” filed on Jun. 20, 2017: U.S. provisional application (“Provisional Application II”), Ser. No. 62/522,661, entitled “3-Dimensional NOR String Arrays in Segmented Stacks,” filed on Jun. 20, 2017; (iii) U.S. provisional application (“Provisional Application III”), Ser. No. 62/522,665, entitled “3-Dimensional NOR String Arrays in Segmented Shared Store Regions,” filed on Jun. 20, 2017; and (iv) U.S. provisional patent application (“Provisional Application IV”), entitled “3-Dimensional NOR Memory Array Architecture and Methods for Fabrication Thereof,” filed on Aug. 25, 2017. The disclosures of the Parent application and the Provisional Applications I-IV are hereby incorporated by reference in their entireties.
This application is also related to U.S. patent application (“non-provisional application”), Ser. No. 15/248,420, entitled “Capacitive-Coupled Non-Volatile Thin-film Transistor Strings in Three-Dimensional Arrays,” filed Aug. 26, 2016. The non-provisional application is hereby incorporated by reference in its entirety. The non-provisional application has been published as U.S. 2017/0092371. References to the non-provisional application herein are made by paragraph numbers of the publication.
BACKGROUND OF THE INVENTION 1. Field of the InventionThe present invention relates to non-volatile NOR-type memory strings. In particular, the present invention relates manufacturing processes for the 3-dimensional structure of such a non-volatile NOR-type memory string.
2. Discussion of the Related ArtIn high density 3-dimensional memory structures, such as those disclosed in the non-provisional application, it is desirable to include a metal sublayer shunt which is electrically connected to either a source sublayer or a drain sublayer. Both sources and drains may be contacted by a conductive sublayer shunt (i.e., as separate conductive sublayers). For example, in the process illustrated in FIG. 5a of the non-provisional application, conductive sublayers may be deposited in addition to source sublayer 521, drain sublayer 523, sacrificial sublayer 522 (which would subsequently be replaced by a channel sublayer). These sublayers are deposited one sublayer at a time and then patterned using photoresist and etched. In this detailed description, the drain, source and channel or sacrificial sublayers, including any associated conductive sublayers, are collectively referred to as the “active layer” and a number of active layers provided one on top of another, separated from one another by a dielectric layer, are referred to as a “NIN stack.”
The metal sublayers are provided to achieve significantly reduced resistance in each of the source and drain sublayers. A lower resistance corresponds to a lower resistance-capacitance (RC) time constant, which results in a higher speed device. For this purpose, it is desirable to achieve low resistance using thick metal-comprising conductive sublayers.
Conductive sublayers having a metal (e.g., tungsten) that can withstand the subsequent elevated temperature processing (>500° C.) are difficult to etch in 3-D memory structures because of etch selectivity. That is, the etch rate of the conductive layer may not be significantly greater than the etch rate of the photoresist and/or hard mask that are used to protect other features that are not to be etched. (In general, to protect the material not intended to be etched, the target material should etch at a significantly faster rate than the masking layer or layers. It would be undesirable that the masking layer or layers are completely removed before etching of the target material is complete.) Etch selectivity becomes an even greater problem as each metal sublayer becomes thicker, as a greater number of metal sublayers are present in the stack (e.g. metal shunt sublayers are provided in both source and drain sublayers), and as more memory layers (e.g., 8 or 16 layers of active strips) are provided. However, to achieve higher density at lower cost, it is desirable to provide 8 or more memory layers.
Another problem encountered in fabrication of these memory structures is their mechanical stability, due to their high aspect ratios. (In this regard, the aspect ratio is the ratio between the structure's height to its width). It has been shown that a semiconductor structure with a high aspect ratio can be mechanically unstable, so that the structure leand or even topples completely during the fabrication process.
SUMMARYThe present invention addresses obtaining low cost, low resistance metal interconnects and mechanical stability in a high aspect ratio structure. According to the various embodiments disclosed herein, the present invention provides a replacement metal process, which defers the need for a metal etching step in the fabrication process until after all patterned photoresist is no longer present. Under this process, the conductive sublayers may be both thick and numerous. The present invention also provides for a strut structure which facilitates etching steps on high aspect ratio structures, which enhances mechanical stability in a high aspect ratio memory stack.
The present invention is better understood upon consideration of the detailed description below, in conjunction with the accompanying drawings.
In this detailed description, like elements in the figures are provided like reference numerals to facilitate reference to features in the figures.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTSVarious embodiments of the present invention are described generally herein. After the various embodiments are described, some specific examples of materials and fabrication steps are described that can be applied to forming the various embodiments.
In this detailed description, the term “patterning” (as applied to a target layer) refers to (i) providing a masking layer (e.g., photoresist layer or hard mask layer) over the target layer, (ii) creating a pattern in the masking layer using suitable photolithography techniques and (iii) transferring the pattern in the masking layer to the target layer using an etching step.
In one example, the masking layer is a “hard mask” layer that is known to those of ordinary skill in the art. To create a masking layer out of a hard mask material, the hard mask material is first provided (e.g., by deposition) on a target layer, over which is then provided a photoresist material. The photoresist pattern is then patterned. The pattern in the photoresist selectively exposes a portion of the hard mask material to an etching agent and protects the remainder of the hard mask material from the etching agent. The pattern of the photoresist is then transferred to the hard mask material by the etching agent, which removes the exposed portion of the hard mask material, so that the protected portion of the hard mask material remains. The hard mask material may then be fixed (e.g., by baking) to become the masking layer for patterning the target layer. After the target layer is patterned, typically by another etching step, the masking layer may be removed in a subsequently step. In examples where a hard mask is not required, the pattern in the photoresist layer may be directly transferred to the target layer. In this detailed description, unless otherwise expressly stated herein, masking materials are removed in due course after completion of the etching step or steps of the target layer.
In this detailed description, methods of fabricating a memory structure over a semiconductor substrate are described. Prior to forming the memory structure, various devices and circuitry are formed on or in the semiconductor substrate using conventional techniques. Examples of methods for connecting bit lines to devices or circuitry on or in the semiconductor substrate are first generally described. Such methods are usually carried out before formation of the memory structures.
Following the description of bit line connections, various embodiments of the present invention relating to the memory structure are described. These embodiments generally relate to various aspects of fabricating a memory structure to form an array of individual memory cells. In these embodiments, the memory structures incorporate the bit lines that have been formed and are connected to the devices and circuitry already formed on or in the semiconductor substrate.
Formation of Connections Between Bit Lines to be Formed in the Memory Structure and Devices in the Semiconductor Substrate—Example 1Alternatively, vias 107 may be etched in the first and second dielectric layers before deposition of sacrificial material SAC 4. In that alternative approach, the SAC4 material also fills vias 107 formed by etching through the first and second dielectric layers.
As mentioned below, sacrificial material SAC4 in sacrificial layer 104 and vias 107 are later replaced simultaneously by a low-resistivity conductor material, such as a metal. Allowing SAC4-filled vias 107 to be later replaced with a low-resistivity metal provides the advantage of significant reduction the resistance in the vertical connectors. Filling vias with N+ doped poly may add resistance to the common drain or bit line, especially for tall NIN stacks. Thus, for tall NIN stacks, filling vias 107 with the SAC4 material for later metal replacement is preferred.
Referring back to
As shown in
In this example, the connections through vias 107 to the 2nd and higher active layers of the memory structure are partially fabricated for each of such active layers during the fabrication of each preceding active layer. Under this scheme, each via is constructed in one or more parts, with each part having a relatively low aspect ratio (relative to the completed via), making fabricating such a via a less challenging effort relative the process described above in conjunction with
Memory Cell Fabrication
Embodiment 1In this embodiment, tall memory structures with large aspect ratios are achieved using reinforcing struts. The high aspect ratio structures are created after the precursor structures are stabilized by a system of strut structures, and one or more sacrificial layers (“SAC4 sublayers”) are removed by one or more etching steps into the exposed sidewalls of trenches through the NIN stack.
In
After deposition of the SAC2 material, the SAC2 material of third sacrificial layer 318 may be removed from the top of each NIN stack, exposing hard mask layer 313. Some SAC2 material may also be removed from the top of the filled trenches, so that the SAC2 material recesses below hard mask layer 313. This partial removal of the SAC2 material may be accomplished using any suitable technique, such as wet or dry etching, CMP, or a combination of such techniques. Strut layer 314 is then deposited. Strut layer 314 may be any suitable material, such as silicon nitride. In some embodiments, strut layer 314 may be provided by the same material or materials as hard mask layer 313. Strut layer 314 is then patterned form struts 314a, 314b and 314c connected the hard mask structures over the NIN stacks.
As shown in
Thereafter, memory structure 300 of
SAC4-containing second sacrificial sublayers 304a and 304b in each active layer of each NIN stack are then removed in whole or in part by etching. This etching may be achieved using a selective chemical etching which does not etch, or etches very little, of the other sublayers in the active layers. After the SAC4 material in second sacrificial sublayers 304a and 304b of each active layer is removed, a conductive material is then deposited to fill in the voids left open by removal of the SAC4 material, thus forming conductive sublayers 319a and 319b. The conductive material also coats the sidewalls of the second set of trenches and the top of the NIN stacks, both of which are then removed by an isotropic or anisotropic etching. Resulting memory structure 300 is shown in
Next, the SAC2 material from trenches 312a, 312b and 312c is removed by selective etching.
Conductive sublayers 319a and 319b may also be sealed to prevent damage from subsequent process steps, such as described below in conjunction with Embodiment 5 described below. Sealing may be achieved after both sides of conductive sublayers 319a and 319b are exposed (i.e., after removal of the SAC2 material from trenches 312a, 312b and 312c), and before channel sublayer 332 and storage layers 335 are formed, as described below.
From memory structure 300 of
Storage layer 335 (e.g., an oxide-nitride-oxide (ONO) layer) is then deposited on memory structure 300 of
As mentioned above, in the example illustrated by
Channel sublayer 332 is formed in this example after conductive or metal layers 319a and 319b replace the SAC4 material in second sacrificial sublayers 309a and 309b (“metal replacement”). In other embodiments, channel sublayer 332 may be formed prior to metal replacement.
In a further example of this embodiment, oxide struts extend the height of the active layers one at a time, rather than using a single mask layer provided to support and extend the height of all the active layers.
After vias 377a, 377b and 377c are etched into active layer 380, ILD material 349 is then deposited to fill the vias. The ILD material is then planarized. The process for forming an active layer, patterning and filling vias 377 with the ILD material are repeated for each active layer.
Oxide struts with tapering cross sections (e.g., pyramid 378 of
Although oxide struts are provided in every trench in
In this embodiment, the SAC4 material-containing sublayer or sublayers (e.g., second sacrificial sublayers 304a and 304b of
Dielectric layer 403 is then deposited over the memory structure of
SAC4 material-containing sublayers of all active layers (e.g., second sacrificial sublayers 304a and 304b) in the NIN stack are then selectively etched. The etching proceeds from each exposed sublayer and extends lengthwise (i.e., in the direction of the lengths of source and drain sublayers 303 and 301) and continues underneath the unexposed portions of storage layer 335, thereby leaving behind long cavities in the SAC4 material-containing sublayers, as the SAC4 material is removed. After removing the SAC4 material-containing sublayers, metal replacement is carried out by depositing a conductive material into the cavities, on the exposed sidewalls of the NIN stacks and on top of the NIN stacks. The conductive material is then removed from the exposed sidewalls of the NIN stacks and from the top of the NIN stacks, thus leaving the conductive sublayer resulting from the cavities being filled. The metal replacement step is then complete.
The NIN stacks of
Alternatively, the NIN stacks can be built incrementally in two or more portions. In each portion, only a limited number of active layers are etched in the vertical direction. Specifically, etching of high aspect ratio NIN stacks are avoided in the initial portion (portion 1) by etching fewer active layers, which reduces the aspect ratio. When a subsequent portion of the NIN stacks is added on top of an earlier portion of the NIN stacks and etched, the earlier portion is supported by a dielectric layer that is deposited in the trenches of the earlier portion. The subsequent portion is self-supporting during its etch, as the etch does not create a high aspect ratio structure. When all the portions of the NIN stacks have been fabricated, a dielectric layer is deposited to fill any remaining open trenches, patterned, and etched to remove all previously deposited like dielectric layers from all earlier portions of the NIN stacks, while maintaining the mechanical strength of the NIN stacks. When the conductive sublayer is inserted into the NIN stacks, only one side of each NIN stack is exposed in a trench, while the opposite side trench is filled with a dielectric layer.
On top of portion 1 of
In some embodiments, portion 1 may be processed through storage layer 335 and local word lines formations, before beginning construction of portion 2.
The conductive sublayers 319a and 319b may also be sealed to protect subsequent steps in the fabrication process. One example of a sealing process is discussed below in conjunction with Embodiment 5. The sealing process may occur after the conductive layers are exposed to the trenches, and before channel and storage sublayer formation.
As explained above, even though channel sublayer 332 may be formed before the metal replacement step (e.g., conductive sublayers 319a and 319b), formation of channel sublayer 332 may also take place after the metal replacement step. Immediately after the metal replacement step, as dielectric layer 509 still occupies every other trench, channel sublayer 332 may be formed only in the active layers exposed to the open trench sides of the NIN stacks. The excess channel material at the bottom of the trench, on the trench sidewalls and at the tops of the NIN stacks are removed by etching. Storage layer 335 (e.g., an ONO layer) may be deposited to line the open trenches and at the tops of the NIN stacks. The excess storage layer material at the bottom of the trenches and on top of the NIN stacks are then removed, leaving storage layer 335 only on the sidewalls of the trenches. Alternatively, the excess storage layer material may be retained until after formation of the word lines.
The conductive material for forming the word lines are then deposited over the storage layer on the side walls of the exposed trenches and patterned to provide the memory structure shown in
The remaining portions of dielectric layer 509 in every other trench are then removed completely to allow formation of storage layer 335 and word line layer 336 using substantially the same process discussed above. The resulting structure is shown in
When each conductive sublayer (e.g., conductive sublayer 319a and 319b, adjacent source sublayer 303 and drain sublayer 301, respectively) includes two or more materials, the resulting cross section would be similar to that of the conductive sublayers 319a and 319b shown in
Further global word lines 106a may be formed above memory structure 500, with vias 109a dropping down to effectuate contact the local word lines. In
Although
In this detailed description, when a storage layer (e.g. an ONO layer) is first formed for a first group of memory cells (e.g., the memory cells of portion 1 or even the memory cells on one side of an active strip) and, subsequently, another storage layer is formed for a second group of memory cells (e.g., the memory cells of portion 2 or the memory cells on the opposite side of the active strip), the storage layer for the first group of memory cells need not be the same as the storage layer for the second group of memory cells. For example, one storage layer may be a relatively thick tunnel dielectric layer (e.g., 5 nanometers or more) to provide a long data retention, albeit slower writes and more limited write/erase cycle endurance, while the other storage layer may be a relatively thin tunnel dielectric layer (e.g., 3 nanometers or less) to provide a short data retention, but faster writes and higher write/erase cycle endurance. As result two or more types of memory cells may be provided in the same memory structure.
Embodiment 4In the non-provisional application, memory cells may be provided only on one side of an NIN stack, but not the other side. Such an arrangement facilities fabrication and eliminates the “cell disturb” problem possible in side-by-side memory cells of the same NIN stack. In this example, after memory cell fabrication is substantially complete (e.g., prior to metal replacement of a sacrificial material, e.g., SAC4 material) and the memory cells are effectively “sealed,” the metal replacement step may then take place to introduce the conductive sublayer into the active layers in each NIN stack, as discussed in conjunction with Embodiment 2 above. In this manner, the deleterious risk of metal contamination in the memory cells is reduced.
Dielectric layer 611 is then deposited over the NIN stacks and into any open trenches. Dielectric layer 611 is then patterned to expose the narrow trenches (e.g., trenches 601-2 and 602-4) and any portions of word line layer 336 over the narrow trenches. The exposed dielectric material 609 in the narrow trenches and the portions of the word line layer 336 over these narrow trenches are then removed by etching.
SAC4 material-containing sublayers (e.g., second sacrificial sublayers 304a and 304b) in the active layers are now exposed in the narrow trenches to allow removal by etching. In this example, the SAC4 material-containing sublayers are not completely removed—e.g., a portion of second sacrificial sublayers 304a and 304b o the far side of the exposed portion (e.g., adjacent to storage sublayer 335) remains. In other examples, second sacrificial sublayer 304a and 304b may be completely removed. Conductive sublayers (e.g., conductive sublayers 319a and 319b, adjacent source and drain sublayers 303 and 301, respectively) are then deposited into the cavities resulting from removing or partially removing the SAC4 material. Excess material from conductive sublayers 319a and 319b on the sidewalls of the trenches and at the top of the NIN stacks is then removed, leaving conductive sublayers 319a and 319b filling the cavities, as shown in
Dielectric layer 612 is then deposited to fill the narrow trenches (e.g., trenches 602-2 and 602-4) and recessed by etching to below the lower surface of the portions of word line layer 336 that sit on top of the NIN stacks. (Alternatively, narrow trenches 602-2 and 602-4 may be left unfilled to serve as air-gap isolation between adjacent NIN stacks; in many application, air-gap isolation is preferred.) Global word lines 106a can then be formed above the NIN stacks using, for example, a dual damascene process, as shown in
In this embodiment, struts which extend the full height of the NIN stacks are fabricated by a via etch and fill method. In addition to an example discussed above in conjunction with
Next, the trenches are filled with third sacrificial layer 318, which contains the SAC2 material discussed above. Resulting structure 700 is then patterned to remove the SAC2 material from every other trench, such as shown in
Next, conductive sublayers 319a and 319b are sealed to avoid cross contaminating memory structure 700 in subsequent processing steps. Sealing may be achieved using a selective etch on conductive sublayers 319a and 319b in each active layer to recess conductive sublayers 319a and 319b from the sidewalls. Dielectric barrier material 712 may be deposited into the recesses, followed by removal of the excess dielectric barrier material 712 from the trenches, leaving dielectric barrier material 712 only in the recesses resulting from selectively etching the conductive sublayers 319a and 319b in each active layer, as shown in
Recessed channel formation then proceeds as described above (i.e., partially removing the SAC1 material from first sacrificial sublayer 302, followed by deposition of channel sublayer 332) to provide resulting structure 700 shown in
Alternatively, protective dielectric sublayer 713 may be provided by a thin 1-5 nm layer of silicon. In some embodiments, protective dielectric layer 713 is not provided. In some embodiments, storage layer 335 and protective dielectric sublayers 713 are removed from neither the top of the NIN stacks nor the floors of the trenches prior to word line formation. In yet other embodiments, storage layers 335 are removed from the floors of only every other trench.
At this point in the process, SAC2 material-containing sacrificial layer 718 may be deposited into the trenches and on top of the NIN stacks. Sacrificial layer 718 may be planarized and removed from hard mask layer 702 at the top of the NIN stacks. Sacrificial layer 718 is then patterned so that vias 719 may be etched into every other trench. Vias 719 extend the full height of the NIN stacks and through underlying dielectric layer 720 to expose the underlying conductors that serve as global word lines, as shown in
Any resulting open trenches as well as the space above the NIN stacks are then filled with deposited dielectric material 721, which may be planarized using a CMP process. Second layer of global word lines 106a may then be formed, after providing vias in the planarized dielectric layer 721 to connect to local word lines in word line layer 336 that are not connected to the global word lines 106s underneath the NIN stacks. Second layer of global word lines 106a may be fabricated, for example, by a dual damascene process. Resulting structure 700 is shown in
While the descriptions for Embodiments 1-4 above do not discuss the steps of fabricating the global word lines 106a and 106s and their connection by vias with the local word lines in local word line layer 336, the description above in conjunction with Embodiment 5 are applicable for forming global word lines 106a and 106s in conjunction with Embodiments 1-4.
Embodiment 6In this embodiment, the spacer hard mask approach is used to pattern the trench features in one module, although the etch of the patterned features may occur in two or more different subsequent modules. This embodiment is advantageous in that no misalignment between trenches, as they are all masked simultaneously. By etching the trenches at two or more different times, high aspect ratio features which may lean or topple are avoided until after the larger structures are stabilized by struts. After strut formation, trench etches that result in high aspect ratio features are performed.
Hard mask features 801 are then selectively removed (e.g., by patterning and etching), leaving behind sidewall features 802 and strut features 804a. Second set of trenches 805 are then etched, using sidewall features 802 as masks, as shown in
In this example, the vertical connections through vias in an ILD layer between the global word lines underneath the memory structure and the local word lines to be formed are fabricated before the memory structure. This embodiment avoids the difficult high aspect ratio via etch of, for example,
The fabrication of a memory structure over the structure of
As described in the Provisional Application III, breaks in the source or drain sublayers 303 and 301 in the NOR strings are introduced to segment such source or drain sublayers horizontally. The segmentation may be achieved by first filling trenches 911-1 to 911-5 with the SAC2 material. After the SAC2 material is planarized to remove excess SAC2 material from the tops of the NIN stacks, memory structure 900 is patterned and etched to create vias in the SAC2 material. Thereafter, the exposed semiconductor sublayers in the active layers (e.g., source sublayer 303, drain sublayer 301 and channel sublayer 332)—but not any of the contacting metal or conductive sublayers 319a and 319b—are selectively removed using, for example, atomic layer etching to cause opens 923 and 921 in the source sublayer 303 and drain sublayer 301 (“segmentation”), respectively. Resulting structure 900 is shown in
After segmentation,
In this detailed description, structs have been used and in various forms to mechanically reinforce where high aspect ratio memory structures are provided. However, in general, when the aspect ratio of the structures is less than about 25, the structures are mechanically stable enough to be free standing without struts or other reinforcement. For example, in a memory structure of 4 NIN stacks, each having a height of about 600 nm and word line spacing of about 30 nm, the aspect ratio is 20 for each NIN stack. In that case, the memory structure, with a smallest feature size of about 30 nm wide, may be fabricated without struts or other reinforcement.
Interlayer Dielectric MaterialInterlayer dielectric materials described in this detail description preferably withstand any etching of the SAC1, SAC2, and SAC4 materials, including when these materials are etched more than once. In choosing an interlayer dielectric material to be used herein, one preferably considers the capacitance that may develops in the NIN stacks. The interlayer dielectric material may preferably be either silicon oxide (e.g., high-temperature oxide (HTO) or another high quality, etch-resistant variety), silicon nitride, or a combination of the two (e.g. partially silicon oxide and partially silicon nitride).
EXAMPLES: SACRIFICIAL MATERIALS AND ETCHESThe SAC1, SAC2, and SAC4 materials may be any suitable sacrificial material, some of which are described in the non-provisional application. Such sacrificial materials include silicon oxide, boron doped silicon oxide (BSG), phosphorus doped silicon oxide (PSG), boron phosphorus doped silicon oxide (BPSG), silicon nitride, silicon carbide, silicon carbon nitride, silicon carbon oxygen hydrogen, germanium, or a combination of some of these materials. The sacrificial layers may be high-density or low-density (i.e., porous), and may be formed using any suitable method, including chemical vapor deposition (CVD), physical vapor deposition (PVD), electrodeposition, sputtering, evaporation, or spin-on techniques. The sacrificial material may be etched by any suitable technique that is selective, i.e. an etch that removes the targeted sacrificial material but does not substantially remove any of the non-targeted layers. For example, hydrofluoric acid (HF) etches SiO2 and variants rapidly, while HF removes SiN and Si at very slow rates.
Example 1: Sacrificial Materials and EtchesThe SAC1 material in this example may include a high temperature silicon oxide (HTO) with a relatively low etch rate in dilute HF. The SAC4 material in this example may include a Ge or a bi-layer of Si and Ge. The SAC4 material may also include BPSG or SiO2 from tetra ethyl orthosilicate (TEOS) with a higher etch rate in dilute HF than the etch rate of HTO. The strut and hard mask materials may include silicon nitride. The SAC2 material in this example may include Ge, which may be etched using hot (70° C.) hydrogen peroxide mixed with water (e.g. 20 vol % H2O2 although any appropriate etch mixture may be used.
As used herein, an “appropriate etch” refers to an etch that etches one material at a rate that is at least 10 times faster than the etch of any and all other materials exposed to the etchant. For example, the hydrogen peroxide wet etch is very selective to Ge (i.e., germanium) and will not etch or minimally etch the other materials (e.g. silicon, silicon dioxide, silicon nitride).
The SAC4 material may be etched in a solution of hydrofluoric (HF) acid and water, or buffered HF. HF or buffered HF will etch BPSG or TEOS at a much faster rate than HTO (e.g., >10:1), and will etch the other materials, silicon and silicon nitride at a much slower rate or not at all.
After removal of the SAC4 and SAC2 materials, the SAC1 material can be partially etched by wet or dry techniques to form the recessed features detailed in the non-provisional application. In some embodiments, the SAC1 “spine” remaining between the two adjacent channels of the same active strip may be removed by selective sideways etching along the length of the active strip to form air-filled cavities, providing the so called “air gap” isolation that has a dielectric constant of 1.0, thereby substantially reducing the parasitic coupling between two adjacent channels.
Example 2: Sacrificial Materials and EtchesThe SAC1 material in this example may include silicon nitride, the SAC4 material in this example may include Ge or a bi-layer of Si and Ge, the SAC2 in this example may include BPSG or TEOS, and the strut and the hard mask in this example may include silicon nitride. Where the SAC4 material includes Ge, it may be etched by hot (70° C.) hydrogen peroxide mixed with water (20 vol % H2O2). This wet etch is very selective to germanium and will not etch or minimally etch the other materials (e.g. silicon, silicon dioxide, silicon nitride). The SAC2 material can be etched in a solution of hydrofluoric (HF) acid and water, or buffered HF which will not etch or minimally etch Ge, Si, or SiN. Finally, after the SAC4 and SAC2 materials have been removed, the SAC1 material can be partially etched using a solution that contains phosphoric acid (wet) or any suitable selective dry etch technique to form the recessed features detailed in the non-provisional application.
Example 3: Sacrificial Materials and EtchesThe SAC1 material in this example may include silicon oxide (HTO) with a relatively low etch rate, the SAC2 material in this example may include BPSG or TEOS with a relatively high etch rate, the SAC4 material in this example may include BPSG or TEOS, and the strut and hard mask may include silicon oxide or silicon nitride. The SAC2 and SAC4 materials may be etched using a solution of hydrofluoric (HF) acid and water, or buffered HF. At the time the SAC2 material is etched, the SAC4 material may be protected from wet etch by photoresist. This wet etch is very selective to BPSG and TEOS and will not etch or minimally etch the other materials (e.g. silicon, HTO, silicon nitride), and will etch HTO (SAC1) slowly. Finally, after the SAC4 and SAC2 materials have been removed, the SAC1 material can be partially etched using wet or dry techniques to form the recessed features detailed in the non-provisional application.
EXAMPLES: CONDUCTIVE SUBLAYER MATERIALSThe conductive sublayers described in this detailed description may be any suitable material or materials, such as titanium, titanium nitride, tungsten nitride, tungsten, titanium tungsten, tantalum, tantalum nitride, cobalt, chrome, molybdenum, or niobium, or combinations or alloys thereof. The metal layer may be deposited using any suitable method, such as CVD, atomic layer deposition (ALD), PVD, sputtering, evaporation, electrodeposition, or any combinations thereof.
The metal layer in this detailed description may be deposited using any suitable method, such as chemical vapor deposition (CVD), atomic layer deposition (ALD), physical vapor deposition (PVD), sputtering, evaporation, electrodeposition, or combinations thereof.
Example 1: Conductive Sublayer MaterialsAn example of a group of sublayers is Ti/TiN/W. The Ti sublayer adheres well to the dielectric or silicon layers, the TiN sublayer is a diffusion barrier, and the W sublayer has lower resistivity than either Ti or TiN. The Ti/TiN layers can be referred to as liner or barrier layers. In general, it is preferred but not required to have the thickness of the liner or the barrier sublayers be less than the low resistivity sublayer. The conductive sublayer may include 1 to 5 nm of titanium, 1 to 5 nm titanium nitride, and 1 to 40 nm of tungsten.
Example 2: Conductive Sublayer MaterialsAnother example of a group of sublayers is TiN/W, wherein the TiN has good adhesion to dielectric or silicon layers and the W sublayer has lower resistivity. The conductive sublayer may include 1 to 5 nm of tungsten nitride and 1 to 40 nm of tungsten.
Other Examples: Conductive Sublayer MaterialsOther groups of sublayers are WN/W, Ta/W, Ta/TaN/W, TaN/W, Ti/Cr, and Ti/TiN/Cr. The examples are not meant to be limiting, and any appropriate combination of sublayers may be utilized. The conductive sub layer may comprise 1 to 5 nm tantalum and 1 to 40 nm tungsten. The metal layer may also include 1 to 5 nm tantalum nitride and 1 to 40 nm tungsten. The metal layer may also include 1 to 40 nm of titanium nitride.
Claims
1. A memory structure, comprising:
- a semiconductor substrate having a substantially planar surface;
- a first stack of active strips and a second stack of active strips formed over the surface of the semiconductor substrate and separated by a predetermined distance along a first direction substantially parallel the planar surface, wherein each stack of active strips comprises two or more active strips provided one on top of another, with adjacent active strips being isolated from each other by an isolation layer, the active strips being substantially aligned lengthwise with each other along a second direction that is substantially also parallel to the planar surface but orthogonal the first direction, and wherein each active strip comprises a dielectric layer provided between a conductive common source layer and a conductive common drain layer, and a channel layer in contact with both the common source layer and the common drain layer, the dielectric layer, the common source layer and the common drain layer being stacked along a third direction that is substantially normal to the planar surface;
- a storage layer; and
- a plurality of conductors, serving as gate electrodes, each extending lengthwise along the third direction, each conductor being within a group of the conductors that are provided between the first stack of active strips and the second stack of active strips and separated from each stack of active strips by the storage layer and the channel layer, thereby forming in each active strip at least one NOR string, each NOR string including a plurality of storage transistors that are formed out of the common source layer, the common drain layer, the channel layer and their adjacent storage layer and the conductors within the group.
2. The memory structure of claim 1, wherein each active strip further comprises at least one metallic layer that is in electrical contact with, and in substantial alignment lengthwise with one of: the common drain layer and the common source layer.
3. The memory structure of claim 1, further comprising a metallic layer contacting one of the second and the third semiconductor layer.
4. The memory structure of claim 3, further comprising:
- a layer of non-conductive material provided between the common drain layer and the isolation layer; and
- a metallic layer that is provided in cavities or recesses formed in the non-conductive layer, the metallic layer being in electrical contact with the common drain layer.
5. The memory structure of claim 1 wherein the non-conductive layer comprises one or more of: silicon oxide, boron doped silicon oxide, phosphorus doped silicon oxide, boron phosphorus doped silicon oxide, silicon nitride, silicon carbide, silicon carbon nitride, silicon carbon oxygen hydrogen, germanium, and any combinations thereof.
6. The memory structure of claim 4 wherein the non-conductive layer is porous.
7. The memory structure of claim 3, wherein the metallic layer further comprises two or more sublayers where a first sublayer is disposed adjacent to and in electrical contact with a second sublayer, and the first sublayer surrounds the second sublayer on three or more sides.
8. The memory structure of claim 7 wherein the thickness of the second sublayer is at least 1.5× the thickness of the first sublayer.
9. The memory structure of claim 3, wherein the metallic layer comprises one or more of: titanium, titanium nitride, tungsten nitride, tungsten, titanium tungsten, tantalum, tantalum nitride, cobalt, chrome, molybdenum, niobium, and any alloys thereof.
10. The memory structure of claim 3, wherein the metallic layer is deposited by atomic layer deposition.
11. The memory structure of claim 1, further comprising a hard mask layer that includes one or more struts formed between two adjacent stacks of active strips, wherein each strut comprises an insulating layer physically connecting the adjacent stacks of active strips.
12. The memory structure of claim 11, wherein the strut is in contact with and disposed to adjacent stacks of active strips at only a portion of the height of one of the stacks of active strips.
13. The memory structure of claim 11, wherein each struct connects adjacent stacks of active strips at the top of the stacks.
14. The memory structure of claim 11, wherein each strut is in contact with and disposed to adjacent stacks of active strips along substantially the entire height of the memory structure.
15. The memory structure of claim 1, wherein the storage layer in the memory structure comprises first and second types of storage material provided at different locations in the memory structure, the first and second types of storage material having different characteristics.
16. The memory structure of claim 15, wherein the first and second types of storage material comprise, respectively, first and second tunnel dielectric layers, the first tunnel dielectric layer being thicker than the second tunnel dielectric layer.
17. The memory structure of claim 15, wherein the first tunnel dielectric layer has a thickness of 5 nm or more.
18. The memory structure of claim 15, wherein the second tunnel dielectric layer has a thickness of 3 nm or less.
19. The memory structure of claim 1, wherein the storage layer comprises an oxide-nitride-oxide material.
4213139 | July 15, 1980 | Rao |
5583808 | December 10, 1996 | Brahmbhatt |
5646886 | July 8, 1997 | Brahmbhatt |
5656842 | August 12, 1997 | Iwamatsu |
5768192 | June 16, 1998 | Eitan |
5789776 | August 4, 1998 | Lancaster et al. |
5915167 | June 22, 1999 | Leedy |
6040605 | March 21, 2000 | Sano et al. |
6107133 | August 22, 2000 | Furukawa et al. |
6118171 | September 12, 2000 | Davies et al. |
6130838 | October 10, 2000 | Kim et al. |
6314046 | November 6, 2001 | Kamiya et al. |
6362508 | March 26, 2002 | Rasovaky et al. |
6434053 | August 13, 2002 | Fujiwara |
6580124 | June 17, 2003 | Cleeves et al. |
6744094 | June 1, 2004 | Forbes |
6774458 | August 10, 2004 | Fricke et al. |
6873004 | March 29, 2005 | Han et al. |
6881994 | April 19, 2005 | Lee et al. |
6946703 | September 20, 2005 | Ryu |
7005350 | February 28, 2006 | Walker |
7223653 | May 29, 2007 | Cheng et al. |
7307308 | December 11, 2007 | Lee |
7489002 | February 10, 2009 | Forbes et al. |
7524725 | April 28, 2009 | Chung |
7542348 | June 2, 2009 | Kim |
7612411 | November 3, 2009 | Walker |
7804145 | September 28, 2010 | Shimizu et al. |
8026521 | September 27, 2011 | Or-Bach et al. |
8139418 | March 20, 2012 | Carman |
8178396 | May 15, 2012 | Sinha et al. |
8237213 | August 7, 2012 | Liu |
8278183 | October 2, 2012 | Lerner |
8395942 | March 12, 2013 | Samachisa et al. |
8630114 | January 14, 2014 | Lue |
8767473 | July 1, 2014 | Shim et al. |
8848425 | September 30, 2014 | Schloss |
8878278 | November 4, 2014 | Alsmeier |
9190293 | November 17, 2015 | Wang et al. |
9202694 | December 1, 2015 | Konevecki et al. |
9230985 | January 5, 2016 | Wu et al. |
9412752 | August 9, 2016 | Yeh et al. |
9455268 | September 27, 2016 | Oh et al. |
9620605 | April 11, 2017 | Liang et al. |
9633944 | April 25, 2017 | Kim |
9748172 | August 29, 2017 | Takaki |
9799761 | October 24, 2017 | Or-Bach et al. |
9842651 | December 12, 2017 | Harari |
9892800 | February 13, 2018 | Harari |
9911497 | March 6, 2018 | Harari |
10074667 | September 11, 2018 | Higashi |
10096364 | October 9, 2018 | Harari |
10121553 | November 6, 2018 | Harari |
10157780 | December 18, 2018 | Wu et al. |
10217667 | February 26, 2019 | Or-Bach et al. |
10249370 | April 2, 2019 | Harari |
10254968 | April 9, 2019 | Gazit et al. |
10283493 | May 7, 2019 | Nishida |
10319696 | June 11, 2019 | Nakano |
10373956 | August 6, 2019 | Gupta et al. |
10381370 | August 13, 2019 | Shin et al. |
10381378 | August 13, 2019 | Harari |
10395737 | August 27, 2019 | Harari |
10431596 | October 1, 2019 | Herner |
10475812 | November 12, 2019 | Harari |
10608008 | March 31, 2020 | Harari |
10608011 | March 31, 2020 | Harari |
10622377 | April 14, 2020 | Harari |
10651153 | May 12, 2020 | Fastow et al. |
10672484 | June 2, 2020 | Cernea |
10692837 | June 23, 2020 | Nguyen et al. |
10692874 | June 23, 2020 | Harari |
10741581 | August 11, 2020 | Harari |
10790023 | September 29, 2020 | Harari |
10818692 | October 27, 2020 | Harari |
10896916 | January 19, 2021 | Harari |
10902917 | January 26, 2021 | Harari |
10950616 | March 16, 2021 | Harari et al. |
11049873 | June 29, 2021 | Hu |
11069696 | July 20, 2021 | Harari |
11120884 | September 14, 2021 | Salahuddin |
11127461 | September 21, 2021 | Harari |
11158620 | October 26, 2021 | Herner |
11180861 | November 23, 2021 | Harari |
11211398 | December 28, 2021 | Herner |
11217600 | January 4, 2022 | Purayath |
11282855 | March 22, 2022 | Herner |
11301172 | April 12, 2022 | Norman |
11309331 | April 19, 2022 | Harari |
11398492 | July 26, 2022 | Yan |
11404431 | August 2, 2022 | Herner |
11507301 | November 22, 2022 | Norman |
11508693 | November 22, 2022 | Norman |
11515309 | November 29, 2022 | Purayath |
11515432 | November 29, 2022 | Salahuddin |
11561911 | January 24, 2023 | Norman |
11580038 | February 14, 2023 | Norman |
20010030340 | October 18, 2001 | Fujiwara |
20010053092 | December 20, 2001 | Kosaka et al. |
20020012271 | January 31, 2002 | Forbes |
20020028541 | March 7, 2002 | Lee et al. |
20020051378 | May 2, 2002 | Ohsawa |
20020193484 | December 19, 2002 | Albee |
20030038318 | February 27, 2003 | Forbes |
20040214387 | October 28, 2004 | Madurawe et al. |
20040246807 | December 9, 2004 | Lee |
20040262681 | December 30, 2004 | Masuoka et al. |
20040262772 | December 30, 2004 | Ramanathan et al. |
20040264247 | December 30, 2004 | Kim |
20050128815 | June 16, 2005 | Ishikawa et al. |
20050280061 | December 22, 2005 | Lee |
20060001083 | January 5, 2006 | Bhattacharyya |
20060155921 | July 13, 2006 | Gorobets et al. |
20070045711 | March 1, 2007 | Bhattacharyya |
20070134876 | June 14, 2007 | Lai et al. |
20080178794 | July 31, 2008 | Cho et al. |
20080239812 | October 2, 2008 | Naofumi et al. |
20080291723 | November 27, 2008 | Wang et al. |
20080301359 | December 4, 2008 | Smith |
20090057722 | March 5, 2009 | Masuoka et al. |
20090157946 | June 18, 2009 | Arya |
20090237996 | September 24, 2009 | Kirsch et al. |
20090279360 | November 12, 2009 | Lee |
20090290442 | November 26, 2009 | Rajan |
20090316487 | December 24, 2009 | Lee et al. |
20100013001 | January 21, 2010 | Cho et al. |
20100121994 | May 13, 2010 | Kim et al. |
20100124116 | May 20, 2010 | Maeda |
20100128509 | May 27, 2010 | Kim et al. |
20100207185 | August 19, 2010 | Lee et al. |
20100219392 | September 2, 2010 | Awaya et al. |
20100254191 | October 7, 2010 | Son et al. |
20110044113 | February 24, 2011 | Kim |
20110047325 | February 24, 2011 | Mishima |
20110115011 | May 19, 2011 | Masuoka et al. |
20110134705 | June 9, 2011 | Jones et al. |
20110170266 | July 14, 2011 | Haensh et al. |
20110208905 | August 25, 2011 | Shaeffer et al. |
20110298013 | December 8, 2011 | Hwang |
20110310683 | December 22, 2011 | Gorobets |
20120063223 | March 15, 2012 | Lee |
20120182801 | July 19, 2012 | Lue |
20120243314 | September 27, 2012 | Maeda |
20120307568 | December 6, 2012 | Banna et al. |
20130031325 | January 31, 2013 | Nakamoto et al. |
20130256780 | October 3, 2013 | Kai et al. |
20140015036 | January 16, 2014 | Fursin et al. |
20140040698 | February 6, 2014 | Loh et al. |
20140075135 | March 13, 2014 | Choi et al. |
20140117366 | May 1, 2014 | Saitoh |
20140151774 | June 5, 2014 | Rhie |
20140229131 | August 14, 2014 | Cohen et al. |
20140247674 | September 4, 2014 | Karda et al. |
20140328128 | November 6, 2014 | Louie et al. |
20140340952 | November 20, 2014 | Ramaswamy et al. |
20150054507 | February 26, 2015 | Gulaka et al. |
20150098272 | April 9, 2015 | Kasolra et al. |
20150113214 | April 23, 2015 | Sutardja |
20150155876 | June 4, 2015 | Jayasena et al. |
20150194440 | July 9, 2015 | Noh et al. |
20150249143 | September 3, 2015 | Sano |
20150263005 | September 17, 2015 | Zhao et al. |
20150372099 | December 24, 2015 | Chen et al. |
20160013156 | January 14, 2016 | Zhai et al. |
20160019951 | January 21, 2016 | Park et al. |
20160035711 | February 4, 2016 | Hu |
20160086970 | March 24, 2016 | Peng |
20160225860 | August 4, 2016 | Karda et al. |
20160300724 | October 13, 2016 | Levy et al. |
20160314042 | October 27, 2016 | Plants |
20170092370 | March 30, 2017 | Harari |
20170092371 | March 30, 2017 | Harari |
20170148517 | May 25, 2017 | Harari |
20170148810 | May 25, 2017 | Kai et al. |
20170213821 | July 27, 2017 | Or-Bach et al. |
20170358594 | December 14, 2017 | Lu et al. |
20180095127 | April 5, 2018 | Pappu et al. |
20180108416 | April 19, 2018 | Harari |
20180269229 | September 20, 2018 | Or-Bach et al. |
20180331042 | November 15, 2018 | Manusharow et al. |
20180366471 | December 20, 2018 | Harari |
20180366485 | December 20, 2018 | Harari |
20180366489 | December 20, 2018 | Harari |
20190006009 | January 3, 2019 | Harari |
20190019564 | January 17, 2019 | Li et al. |
20190067327 | February 28, 2019 | Herner et al. |
20190157296 | May 23, 2019 | Harari et al. |
20190180821 | June 13, 2019 | Harari |
20190206890 | July 4, 2019 | Harari |
20190214077 | July 11, 2019 | Oh et al. |
20190238134 | August 1, 2019 | Lee et al. |
20190244971 | August 8, 2019 | Harari |
20190259769 | August 22, 2019 | Karda et al. |
20190303042 | October 3, 2019 | Kim et al. |
20190319044 | October 17, 2019 | Harari |
20190325945 | October 24, 2019 | Linus |
20190325964 | October 24, 2019 | Harari |
20190348424 | November 14, 2019 | Karda et al. |
20190355747 | November 21, 2019 | Herner |
20190370117 | December 5, 2019 | Fruchtman et al. |
20200051990 | February 13, 2020 | Harari |
20200098738 | March 26, 2020 | Herner et al. |
20200098779 | March 26, 2020 | Cernea |
20200176468 | June 4, 2020 | Herner |
20200258897 | August 13, 2020 | Yan et al. |
20200258903 | August 13, 2020 | Harari |
20200403002 | December 24, 2020 | Harari |
20220238545 | July 28, 2022 | Harari |
2010108522 | May 2010 | JP |
2011028540 | February 2011 | JP |
7203054 | January 2023 | JP |
20120085591 | August 2012 | KR |
20120085603 | August 2012 | KR |
2018236937 | December 2018 | WO |
WO-2018236937 | December 2018 | WO |
- “EP Extended Search Report EP168690149.3”, dated Oct. 18, 2019.
- “European Search Report, EP 16852238.1”, dated Mar. 28, 2019.
- “European Search Report, EP17844550.8”, dated Aug. 12, 2020, 11 pages.
- “Invitation to Pay Additional Fees (PCT/ISA/206), PCT/US2020/015710”, dated Mar. 20, 2020, 2 pages.
- “Notification of Reasons for Refusal, Japanese Patent Application 2018-527740”, (English translation), dated Nov. 4, 2020, 8 pages.
- “Partial European Search Report EP 16869049.3”, dated Jul. 1, 2019, pp. 1-12.
- “PCT Search Report and Written Opinion, PCT/US2018/038373”, dated Sep. 10, 2018.
- “PCT Search Report and Written Opinion, PCT/US2019/014319”, dated Apr. 15, 2019.
- “PCT Search Report and Written Opinion, PCT/US2019/052164”, dated Feb. 27, 2020.
- “PCT Search Report and Written Opinion, PCT/US2019/052446”, dated Dec. 11, 2019.
- “PCT Search Report and Written Opinion, PCT/US2020/015710”, dated Jun. 9, 2020.
- “PCT Search Report and Written Opinion, PCT/US2020/017494”, dated Jul. 20, 2020, 13 pages.
- “PCT Search Report and Written Opinion, PCT/US2020/065374”, dated Mar. 15, 2021, 17 pages.
- “PCT Search Report and Written Opinion, PCT/US2020/065670”, dated Apr. 5, 2021, 12 pages.
- “PCT Search Report and Written Opinion, PCT/US2021/016964”, dated Jun. 15, 2021, 19 pages.
- “PCT Search Report and Written Opinion, PCT/US2021/025722”, dated Jun. 15, 2021, 10 pages.
- Hou, S. Y., et al., “Wafer-Leval Integration of an Advanced Logic-Memory System Through the Second-Generation CoWoS Technology”, IEEE Transactions on Electron Devices, vol. 64, No. 10, Oct. 2017, 4071-4077.
- Kim, N. , et al., “Multi-layered Vertical gate NANO Flash Overcoming Stacking Limit for Terabit Density Storage”, Symposium on VLSI Tech. Dig. of Technical Papers, 2009, pp. 188-189.
- Lue, H.T. , et al., “A Highly Scalable 8-Layer 3D Vertical-gate {VG) TFT Nano Flash Using Junction-Free Buried Channel BE-SONOS Device”, Symposium on VLSI: Tech. Dig. Of Technical Papers, 2010, pp. 131-132.
- Tanaka, T. , et al., “A 768 GB 3b/cell 3D-Floaling-Gate NANO Flash Memory”, Digest of Technical Papers, the 2016 EEE International Solid-Slate Circuits Conference, 2016, pp. 142-144.
- Wann, H.C. , et al., “High-Endurance Ultra-Thin Tunnel Oxide in Monos Device Structure for Dynamic Memory Application”, IEEE Electron Device letters, vol. 16, No. 11, Nov. 1995, pp. 491-493.
Type: Grant
Filed: Mar 9, 2022
Date of Patent: Aug 15, 2023
Patent Publication Number: 20220199643
Assignee: SunRise Memory Corporation (San Jose, CA)
Inventors: Eli Harari (Saratoga, CA), Scott Brad Herner (Portland, OR), Wu-Yi Henry Chien (San Jose, CA)
Primary Examiner: Laura M Menz
Application Number: 17/690,943
International Classification: H01L 21/00 (20060101); H10B 43/27 (20230101); H01L 21/768 (20060101); H01L 23/00 (20060101); H10B 43/20 (20230101); H01L 21/311 (20060101); G11C 16/04 (20060101);